September 2015

Overview

Questions from last week?

  • dplyr: data wrangling
  • tidyr: data reformating

Homeworks posted, due next week.

Today

  • Definitions, motives, history, spectrum
  • Current practices
  • A selection of tools to improve reproducibility
  • Challenges, standards & our role in the future of reproducible research
  • Example markdown document and Git!

"The goal of reproducible research is to tie specific instructions to data analysis and experimental data so that scholarship can be recreated, better understood and verified." Max Kuhn, CRAN Task View: Reproducible Research

History of reproducible research

  • Mathematics (400 BC?)
  • Write scientific paper, Galileo, Pasteur, etc. (1660s?)
  • Publish a pidgin algorithm and describe simulation datasets (1950s?)
  • Sell magtape of code and data (1970s?)
  • Place idiosyncratic dataset & software at website (1990s?)
  • Publish datasets and scripts at website, eg. biology, political science, genetics, statistics (2000s?)
  • Hosted integrated code and data (2020s?)

Gavish & Gonoho AAAS 2011, Oxberry 2013

Motivations:
Claerbout's principle

"An article about computational result is advertising, not scholarship. The actual scholarship is the full software environment, code and data, that produced the result." Claerbout and Karrenbach, Proceedings of the 62nd Annual International Meeting of the Society of Exploration Geophysics. 1992

Benefits are straightforward

  • Verification & Reliability: Find and fix bugs. Today's results == tomorrow's.
  • Transparency: increased citation count, broader impact, improved institutional memory
  • Efficiency: Reduces duplication of effort. Payoff in the (not so) long run
  • Flexibility: When you don’t 'point-and-click' you gain many new analytic options.

But limitations are substantial

Technical

  • Classified/sensitive/big data
  • Nondisclosure agreements & intellectual property
  • Software licensing issues
  • Competition
  • Neither necessary nor sufficient for correctness (but useful in disputes)

Cultural & personal

  • Few follow even minimal reproducibility standards.
  • Few expect reproducibility
  • No uniform standards
  • Inertia & embarassment

Our work exists on a spectrum of reproducibility

alt text

Peng 2011, Science 334(6060) pp. 1226-1227

Goal: expose more of the research workflow

Common practices of many scientists

  • Enter data in Excel
  • Use Excel for data cleaning & descriptive statistics
  • Import data into SPSS/SAS/Stata for further analysis
  • Use point-and-click options to run statistical analyses
  • Copy & paste output to Word document, repeatedly







Common practices of many scientists

  • Version control is ad hoc
  • Excel handles missing data inconsistently and sometimes incorrectly
  • Excel uses poor algorithms for many functions
  • Scripting is possible but rare











alt text

Click trails are ephemeral & dangerous

  • Lots of human effort for tedious & time-wasting tasks
  • Error-prone due to manual & ad hoc data handling
  • Difficult to record - hard to reconstruct a 'click history'
  • Tiny changes in data or method require extensive reworking





alt text

Case study: Reinhart and Rogoff controversy

alt text

  • 2010: Claimed high debt-to-GDP ratios led to low GDP growth
  • Threshold to growth at a debt-to-GDP ratio of >90%
  • Substantial popular impact on autsterity politics

Case study: Reinhart and Rogoff controversy

alt text

A coding error in their Excel spreadsheet sliced several countries out of the data set…. The Economist

Case study: Gay Marrage

In May 2015 Science retracted a study of how canvassers can sway people's opinions about gay marriage published just 5 months ago.

  • Original survey data not made available for independent reproduction of results
  • Survey incentives misrepresented
  • Sponsorship statement false.

- Science Editor-in-Chief Marcia McNutt

Two Berkeley grad students who attempted to replicate the study quickly discovered that the data must have been faked.

Source

Case study: Seizure Medicine

2013 Seizure study retracted after authors realize data got "terribly mixed"

"The article has been retracted at the request of the authors. After carefully re-examining the data presented in the article, they identified that data of two different hospitals got terribly mixed. The published results cannot be reproduced in accordance with scientific and clinical correctness."" Authors of Low Dose Lidocaine for Refractory Seizures in Preterm Neonates

Source

Bad spreadsheet merge kills depression paper, quick fix resurrects it

Authors informed the journal that the merge of lab results and other survey data used in the paper resulted in an error regarding the identification codes. Results of the analyses were based on the data set in which this error occurred.

"Lower levels of CSF IL-6 were associated with current depression and with future depression […]" Original conclusion

"Higher levels of CSF IL-6 and IL-8 were associated with current depression […]" Revised conclusion

Source

Scripted analyses are superior

alt text

  • Plain text files readable for a long time
  • Improved transparency, automation, maintanability, accessibility, standardisation, modularity, portability, efficiency, communicability of process (what more could we want?)
  • Steeper learning curve

Literate statistical programming

"Instead of imagining that our main task is to instruct a computer what to do, let us concentrate rather on explaining to humans what we want the computer to do." Donald E. Knuth, Literate Programming, 1984

For example… Let's calculate the current time in R.

time <- format(Sys.time(), "%a %d %b %X %Y")

The text and R code are interwoven in the output:

The time is `r time`

The time is Mon 21 Sep 16:49:00 2015

Literate programming: for and against

For

  • Text & code in one place, in logical order
  • Tables and figures automatically updated
  • Automatic test when building document

Against

  • Text and code in one place; can be hard to read
  • Can slow down the processing of documents (use caching!)

Reproducible Research in R

Need a programming language

The machine-readable part: R

  • R: Free, open source, cross-platform, highly interactive, huge user community in academica and private sector
  • R packages an ideal 'Compendium'?
  • Scriptability → R
  • Literate programming → R Markdown
  • Version control → Git / GitHub

alt text

"both a container for the different elements that make up the document and its computations (i.e. text, code, data, etc.), and as a means for distributing, managing and updating the collection… allow us to move from an era of advertisement to one where our scholarship itself is published" Gentleman and Temple Lang 2004

Very low barrier to documentation of code with roxygen2

alt text

Need a document formatting language

alt text

Markdown: lightweight document formatting syntax. Easy to write, read and publish as-is.

The human-readable part

rmarkdown: - minor extensions to allow R code display and execution - embed images in html files (convenient for sharing) - equations

e.g.

  • * for bullet
  • _ for underline

Dynamic documents in R

knitr - descendant of Sweave

Engine for dynamic report generation in R

alt text


  • Narrative and code in the same file or explicitly linked
  • When data or narrative are updated, the document is automatically updated
  • Data treated as 'read only'
  • Output treated as disposable

Pandoc: document converter

Version Control

Tracking changes with version control

Payoffs

  • Eases collaboration
  • Can track changes in any file type (ideally plain text)
  • Can revert file to any point in its tracked history

Costs - Learning curve

alt text alt text alt text

Environment for reproducible research

alt text

  • integrated R console
  • deep support for markdown and git
  • package development tools, etc. etc.

RStudio 'projects' make version control & document preparation simple

Depositing code and data

Payoffs - Free space for hosting (and paid options) - Assignment of persistent DOIs - Tracking citation metrics

Costs - Sometimes license restrictions (CC-BY & CC0) - Limited or no private storage space

alt text alt text alt text

A hierarchy of reproducibility

  • Good: Use code with an integrated development environment (IDE). Minimize pointing and clicking (RStudio)
  • Better: Use version control. Help yourself keep track of changes, fix bugs and improve project management (RStudio & Git & GitHub or BitBucket)
  • Best: Use embedded narrative and code to explicitly link code, text and data, save yourself time, save reviewers time, improve your code. (RStudio & Git & GitHub or BitBucket & rmarkdown & knitr & data repository)

alt text Stodden (IASSIST 2010) sampled American academics registered at the Machine Learning conference NIPS (134 responses from 593 requests (23%). Red = communitarian norms, Blue = private incentives

alt text Stodden (IASSIST 2010) sampled American academics registered at the Machine Learning conference NIPS (134 responses from 593 requests (23%). Red = communitarian norms, Blue = private incentives

Standards to normalise reproducible research

Reproducible Research Standard (Stodden 2009)

  • The full compendium on the internet
  • Media such as text, figures, tables with Creative Commons Attribution license (CC-BY)
  • Code with one of Apache 2.0, MIT, LGPL, BSD, etc.
  • Original "selection and arrangement" of data with CC0 or CC-BY

Biggest challenge: culture change

Promote culture change through positive attribution

Implement mechanisms to indicate & encourage degrees of compliance (ie. clear definitions for different levels of reproducibility), cf. Stodden's:

  • 'Reproducible': compendium of text-code-data online
  • 'Reproduced': compendium available and independently reproduced
  • 'Semi-Reproducible': when the full compendium is not released
  • 'Semi-Reproduced': independent reproduction with other data
  • 'Perpetually Reproducible': streaming data

Our role in the future of reproducible research (Leveque et al 2012)

  • Train students by putting homework, assignments & dissertations on the reproducible research spectrum
  • Publish examples of reproducible research in our field
  • Request code & data when reviewing
  • Submit to & review for journals that support reproducible research
  • Critically review & audit data management plans in grant proposals
  • Consider reproducibility wherever possible in hiring, promotion & reference letters.

Demo: let's get started

Install Git on your computer

Creating an account on GitHub

  1. Create a GitHub account at https://github.com/
    • This will be a public account associated with your name
    • Choose a username wisely for future use
    • Don't worry about details, you can fill them in later
  2. Create a repository called demo
    • Add a brief and informative description
    • Choose "Public"
    • Check the box for "Initialize this repository with a README"
  3. Click "Create Repository"

In RStudio: clone the repository

  1. Go to RStudio
    • File -> New Project
    • Version Control: Checkout a project from a version control repository
    • Git: Clone a project from a repository
  2. Fill in the info:
    • URL: use HTTPS address
    • Create as a subdirectory of: Browse and create a new folder called GEO503 (or similar)

R Markdown

Create new file

File -> New File -> RMarkdown -> Document -> HTML

alt text

Step 1: Load packages

All R code to be run must be in a code chunk like this:

#```{r,eval=F}
CODE HERE
#```

Load these packages in a code chunk:

library(dplyr)
library(ggplot2)
library(maps)
library(spocc)

Do you think you should put install.packages() calls in your script?

Step 2: Load data

Now use the occ() function to download all the occurrence records for the American robin (Turdus migratorius) from the Global Biodiversity Information Facility.

alt text

Licensed under CC BY-SA 3.0 via Wikimedia Commons

## define which species to query
sp='Turdus migratorius'

## run the query and convert to data.frame()
d = occ(query=sp, from='ebird',limit = 1000) %>% occ2df()

This can take a few seconds.

Step 3: Map it

# Load coastline
map=map_data("world")

ggplot(d,aes(x=longitude,y=latitude))+
  geom_polygon(aes(x=long,y=lat,group=group,order=order),data=map)+
  geom_point(col="red")+
  coord_equal()

Step 4:

Update the YAML header to keep the markdown file

From this:

title: "Untitled"
author: "Adam M. Wilson"
date: "September 21, 2015"
output: html_document

To this:

title: "Demo"
author: "Adam M. Wilson"
date: "September 21, 2015"
output: 
  html_document:
      keep_md: true

And click knit HTML to generate the output

Step 5: Commit and push all changes to GitHub

  1. Stage
  2. Commit (with a message)
  3. Push

Step 6: Explore markdown functions

  1. Use the Cheatsheet to add sections and some example narrative.
  2. Try changing changing the species name to your favorite species and re-run the report.
  3. Stage, Commit, Push!
  4. Explore the markdown file on the GitHub website.

alt text

Final

Abandoning the habit of secrecy in favor of process transparency and peer review was the crucial step by which alchemy became chemistry.Raymond, E. S., 2004, The art of UNIX programming: Addison-Wesley.

Colophon

References

See Rmd file for full references and sources